The advantages and disadvantages of using pre or post stratification methods in caries clinical trials were assessed. It was shown that provided one has knowledge of one or more variables which are correlated with caries increments, the advantages of using a stratification method dominate, and therefore, stratification methods ought to be used. In those instances, not that commonly encountered, where the relationship can be assumed to be linear, the analysis of covariance performed just as well. It was also shown that provided the imbalance among cells is not too severe, both the additive and interaction models were robust, i.e., performed well even though the underlying assumptions needed for such analyses were unmet.